From dhilvert at gmail.com  Fri Jun 13 16:49:40 2008
From: dhilvert at gmail.com (David Hilvert)
Date: Fri Jun 13 16:50:01 2008
Subject: [ale] 0.9.0.1
Message-ID: <20080613184940.e0415969.dhilvert@gmail.com>

Testing release 0.9.0.1 is now available for download.

URLs:

   http://auricle.dyndns.org/ALE/download/ale-0.9.0.1.tar.gz
   http://auricle.dyndns.org/ALE/download/ale-0_9_0_1-win32.zip

Overview:

   This testing release adds support for regional adjustment of tone (e.g.,
   for varying lighting conditions), includes various multi-alignment fixes
   and updates, and adds a man page.

Changelog:

     o Use a pager (e.g., less) when displaying help.
     o Add a man page, at the recommendation of Debian developer Ruben
       Molina. The man page structure roughly follows that of Debian
       Iceweasel, GNU tar, and the earlier ALE manual page for Debian by Nick
       Rusnov and Ruben Molina.
     o Add new algorithms for pixel assignment in multi-alignment, including
       filling, least-size local selection, and global-best selection.
     o Fix various multi-alignment bugs, including a bug causing increased
       computational complexity for rendering, as well as offset and scaling
       bugs.
     o Modify --lod-max to --lod-preferred, add a dimension restriction
       --min-dimension on level-of-detail, and make --lod-preferred=-3 and
       --min-dimension=10 default.
     o Revise DejaGNU test suite for current option syntax, so that "make
       check" now runs all tests.
     o Do not attempt to handle --trans-{save,load} arguments as images in
       the ale wrapper script.
     o Add include <string.h> to ui/ui.h in an attempt to resolve Debian bug
       #482765, reported by Marc Brockschmidt and noted by Ruben Molina.
     o Include more information about scale cluster structures in the logging
       ui.
     o Add tonal registration by element to multi-alignment code.
     o Add a minimum average certainty for registration of multi-alignment
       elements.
     o Modify ale script to tally thread statistics when the environment
       variable ALE_COUNT_THREADS is set non-zero.


From dhilvert at gmail.com  Fri Jun 13 17:02:33 2008
From: dhilvert at gmail.com (David Hilvert)
Date: Fri Jun 13 17:02:48 2008
Subject: [ale] Win32 binaries were linked with ImageMagick;
 now rebuilt without ImageMagick
Message-ID: <20080613190233.feceb0dc.dhilvert@gmail.com>

While testing the 0.9.0.1 downloads, I noticed that the Win32 build was
reporting having been built with ImageMagick, which came as somewhat of a
surprise.  On further inspection, older builds for 0.9.0 and 0.8.11 had also
been linked with ImageMagick, apparently as a result of some testing I had done
earlier on whether auto-configuration with ImageMagick would work under
cross-compilation for Win32.

As no one complained, I assume that these builds worked correctly, so they may
eventually be restored at some point; for how, however, I have updated the
builds to be PPM-only, as advertised.  If you have a copy of the
ImageMagick-linked build, please review the license terms of ImageMagick to
ensure that these are being followed.  Apologies for any inconvenience caused
by the error.

From dhilvert at gmail.com  Sun Jun 15 02:55:11 2008
From: dhilvert at gmail.com (David Hilvert)
Date: Sun Jun 15 02:55:31 2008
Subject: [ale] Variable lighting results
Message-ID: <20080615045511.b9237919.dhilvert@gmail.com>

The variable lighting sets by Gregory Maxwell and Mika Yrj?l? were quite
challenging to work with; these sets motivated --multi=fill (now default) over
--multi=local.  While the current code still does not produce outstanding
results on these sets, the results provide some idea of the difference between
using a single, global tonal registration area and using multiple local tonal
registration areas (elements).


Detail of results from one of Mika Yrj?l?'s sets:

http://auricle.dyndns.org/ALE/tone/


Detail of results from Gregory Maxwell's viola set:

http://auricle.dyndns.org/ALE/tone2/


(Note that, while --md 0 was used to disable local registration in the second
set, this no longer works in 0.9.0.1; --multi=none should be used, instead.)